How does one think about the “off-diagonal” part of the R-matrix?

The universal R-matrix of a quantized universal enveloping algebra is typically written as the product of two terms, one only involving elements of the Cartan, and one only involving elements of the upper and lower triangular pieces of $U_q(g)$.

How do I think about commuting these individual pieces past things like coproducts? Are calculations like this written somewhere?

-